opensource/RSSHub
3
0
Fork 0
mirror of https://github.com/DIYgod/RSSHub.git synced 2026-09-24 23:12:34 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat(route): 观察者网输出全文 (#9338)

Browse Source 
* feature(route): 观察者网输出全文

* fix: add radar

* fix typo

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* fix: add timezone

Co-authored-by: Tony <TonyRL@users.noreply.github.com>
This commit is contained in:
Ethan Shen
2022-03-19 21:01:03 +08:00
committed by GitHub
co-authored by Tony
parent b2137fc283
commit 6477e6b8e4
10 changed files with 161 additions and 66 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/new-media.md
+2 -2
View File
@@ -1869,7 +1869,7 @@ area 分区选项
### 首页
<Route author="nczitzk Jeason0228" example="/guancha" path="/guancha/:caty?" :paramsDesc="['分类，见下表，默认为全部']">
<Route author="nczitzk Jeason0228" example="/guancha" path="/guancha/:category?" :paramsDesc="['分类，见下表，默认为全部']">
| 全部 | 评论 & 研究 | 要闻 | 风闻 | 热点新闻 | 滚动新闻 |
| --- | ------- | ----- | ------- | ------ | ------- |
@@ -1889,7 +1889,7 @@ others = 热点新闻 + 滚动新闻
### 观学院
<Route author="nczitzk" example="/guancha/member/recommend" path="/guancha/member/:caty?" :paramsDesc="['分类，见下表']">
<Route author="nczitzk" example="/guancha/member/recommend" path="/guancha/member/:category?" :paramsDesc="['分类，见下表']">
| 精选 | 观书堂 | 在线课 | 观学院 |
| --------- | ----- | ------- | -------- |
lib/router.js
+8 -8
View File
@@ -1624,15 +1624,15 @@ router.get('/zfrontier/postlist/:type', lazyloadRouteHandler('./routes/zfrontier
router.get('/zfrontier/board/:boardId', lazyloadRouteHandler('./routes/zfrontier/board_postlist'));
// 观察者网
router.get('/guancha/headline', lazyloadRouteHandler('./routes/guancha/headline'));
router.get('/guancha/topic/:id/:order?', lazyloadRouteHandler('./routes/guancha/topic'));
router.get('/guancha/member/:caty?', lazyloadRouteHandler('./routes/guancha/member'));
router.get('/guancha/personalpage/:uid', lazyloadRouteHandler('./routes/guancha/personalpage'));
router.get('/guancha/:caty?', lazyloadRouteHandler('./routes/guancha/index'));
// router.get('/guancha/headline', lazyloadRouteHandler('./routes/guancha/headline'));
// router.get('/guancha/topic/:id/:order?', lazyloadRouteHandler('./routes/guancha/topic'));
// router.get('/guancha/member/:caty?', lazyloadRouteHandler('./routes/guancha/member'));
// router.get('/guancha/personalpage/:uid', lazyloadRouteHandler('./routes/guancha/personalpage'));
// router.get('/guancha/:caty?', lazyloadRouteHandler('./routes/guancha/index'));
router.get('/guanchazhe/topic/:id/:order?', lazyloadRouteHandler('./routes/guancha/topic'));
router.get('/guanchazhe/personalpage/:uid', lazyloadRouteHandler('./routes/guancha/personalpage'));
router.get('/guanchazhe/index/:caty?', lazyloadRouteHandler('./routes/guancha/index'));
// router.get('/guanchazhe/topic/:id/:order?', lazyloadRouteHandler('./routes/guancha/topic'));
// router.get('/guanchazhe/personalpage/:uid', lazyloadRouteHandler('./routes/guancha/personalpage'));
// router.get('/guanchazhe/index/:caty?', lazyloadRouteHandler('./routes/guancha/index'));
// Hpoi 手办维基
router.get('/hpoi/info/:type?', lazyloadRouteHandler('./routes/hpoi/info'));
lib/routes/guancha/headline.js → lib/v2/guancha/headline.js
+11 -8
View File
@@ -1,9 +1,12 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const timezone = require('@/utils/timezone');
const { parseDate } = require('@/utils/parse-date');
module.exports = async (ctx) => {
const rootUrl = 'https://www.guancha.cn';
const currentUrl = `${rootUrl}/GuanChaZheTouTiao/list_1.shtml`;
const response = await got({
method: 'get',
url: currentUrl,
@@ -11,20 +14,20 @@ module.exports = async (ctx) => {
const $ = cheerio.load(response.data);
const list = $('.headline-list li .content-headline h3 a')
.map((_, item) => {
let items = $('.headline-list li .content-headline h3 a')
.toArray()
.map((item) => {
item = $(item);
return {
title: item.text(),
link: `${rootUrl}${item.attr('href')}`,
description: item.parent().next().html(),
pubDate: new Date(item.parents('div').eq(0).find('span').text()).toUTCString(),
link: `${rootUrl}${item.attr('href').replace(/\.shtml$/, '_s.shtml')}`,
pubDate: timezone(parseDate(item.parents('div').first().find('span').text()), +8),
};
})
.get();
});
const items = await Promise.all(
list.map((item) =>
items = await Promise.all(
items.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const detailResponse = await got({
method: 'get',
lib/routes/guancha/index.js → lib/v2/guancha/index.js
+55 -28
View File
@@ -1,21 +1,42 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const date = require('@/utils/date');
const timezone = require('@/utils/timezone');
const { parseDate } = require('@/utils/parse-date');
const { parseRelativeDate } = require('@/utils/parse-date');
const config = {
review: { title: '评论 & 研究', query: '.module-news-main' },
story: { title: '要闻', query: '.img-List' },
fengwen: { title: '风闻', query: '.fengwen-list' },
redian: { title: '热点' },
gundong: { title: '滚动' },
all: { title: '全部' },
home: { title: '首页' },
others: { title: '热点 & 滚动' },
review: {
title: '评论 & 研究',
query: '.module-news-main',
},
story: {
title: '要闻',
query: '.img-List',
},
fengwen: {
title: '风闻',
query: '.fengwen-list',
},
redian: {
title: '热点',
},
gundong: {
title: '滚动',
},
all: {
title: '全部',
},
home: {
title: '首页',
},
others: {
title: '热点 & 滚动',
},
};
module.exports = async (ctx) => {
const total = 10;
const caty = ctx.params.caty || 'all';
const category = ctx.params.category ?? 'all';
const rootUrl = 'https://www.guancha.cn';
let newsList = [],
@@ -24,7 +45,7 @@ module.exports = async (ctx) => {
// 'review', 'story' and 'fengwen' come from homepage.
if (caty === 'review' || caty === 'story' || caty === 'fengwen' || caty === 'all' || caty === 'home') {
if (category === 'review' || category === 'story' || category === 'fengwen' || category === 'all' || category === 'home') {
const response = await got({
method: 'get',
url: rootUrl,
@@ -35,54 +56,59 @@ module.exports = async (ctx) => {
const fetchPost = (slice) =>
slice
.find('h4.module-title a')
.toArray()
// Filter some blank links which lead to no contents but 'https://user.guancha.cn'.
.filter((_, item) => $(item).attr('href') !== 'https://user.guancha.cn')
.map((_, item) => {
.filter((item) => $(item).attr('href') !== 'https://user.guancha.cn')
.map((item) => {
item = $(item);
const link = item.attr('href');
return {
title: item.text(),
link: (link.indexOf('http') === 0 ? '' : rootUrl) + link,
link: `${link.indexOf('http') === 0 ? '' : rootUrl}${link.replace(/\.shtml/, '_s.shtml')}`,
};
})
.get();
});
if (caty === 'all' || caty === 'home') {
if (category === 'all' || category === 'home') {
newsList = fetchPost($(config.review.query))
.slice(0, total / 3)
.concat(fetchPost($(config.story.query)).slice(0, total / 3), fetchPost($(config.fengwen.query)).slice(0, total / 3));
} else {
newsList = fetchPost($(config[caty].query)).slice(0, total);
newsList = fetchPost($(config[category].query)).slice(0, total);
}
}
// 'redian' and 'gundong' come from api.
if (caty === 'redian' || caty === 'all' || caty === 'others') {
if (category === 'redian' || category === 'all' || category === 'others') {
const response = await got({
method: 'get',
url: `${rootUrl}/api/redian.htm`,
});
redianList = response.data.items
.map((item) => ({
title: item.TITLE,
link: `${rootUrl}${item.HTTP_URL}`,
link: `${rootUrl}${item.HTTP_URL.replace(/\.shtml/, '_s.shtml')}`,
}))
.slice(0, caty === 'all' ? total / 3 : total);
.slice(0, category === 'all' ? total / 3 : total);
}
if (caty === 'gundong' || caty === 'all' || caty === 'others') {
if (category === 'gundong' || category === 'all' || category === 'others') {
const response = await got({
method: 'get',
url: `${rootUrl}/api/gundong.htm`,
});
gundongList = response.data.items
.map((item) => ({
title: item.TITLE,
link: `${rootUrl}${item.HTTP_URL}`,
link: `${rootUrl}${item.HTTP_URL.replace(/\.shtml/, '_s.shtml')}`,
}))
.slice(0, caty === 'all' ? total / 3 : total);
.slice(0, category === 'all' ? total / 3 : total);
}
const items = await Promise.all(
@@ -98,8 +124,9 @@ module.exports = async (ctx) => {
// => https://user.guancha.cn/main/content?id=399176
const jumpMatch = detailResponse.data.match(/user.guancha.cn\/main\/content\?id=(.*)";/);
if (jumpMatch !== null) {
item.link = `https://user.guancha.cn/main/content?id=${jumpMatch[1]}`;
item.link = `https://user.guancha.cn/main/content?id=${jumpMatch[1]}&page=0`;
detailResponse = await got({
method: 'get',
url: item.link,
@@ -112,9 +139,9 @@ module.exports = async (ctx) => {
if (dateMatch === null) {
// PubDates of posts in 'fengwen' are in an informal format.
item.pubDate = date(content('.time1').text(), 8);
item.pubDate = parseRelativeDate(content('.time1').text());
} else {
item.pubDate = new Date(dateMatch[1]).toUTCString();
item.pubDate = timezone(parseDate(dateMatch[1]), +8);
}
item.description = content('.all-txt').html() || content('.article-txt-content').html();
@@ -125,7 +152,7 @@ module.exports = async (ctx) => {
);
ctx.state.data = {
title: `观察者网 - ${config[caty].title}`,
title: `观察者网 - ${config[category].title}`,
link: rootUrl,
item: items,
};
lib/v2/guancha/maintainer.js
+7
View File
@@ -0,0 +1,7 @@
module.exports = {
'/:category?': ['nczitzk', 'Jeason0228'],
'/headline': ['nczitzk'],
'/member/:category?': ['nczitzk'],
'/personalpage/:uid': ['Jeason0228'],
'/topic/:id/:order?': ['occupy5', 'nczitzk'],
};
lib/routes/guancha/member.js → lib/v2/guancha/member.js
+8 -7
View File
@@ -1,8 +1,8 @@
const got = require('@/utils/got');
const { parseDate } = require('@/utils/parse-date');
const timezone = require('@/utils/timezone');
const { parseDate } = require('@/utils/parse-date');
const title = {
const titles = {
recommend: '精选',
books: '观书堂',
courses: '在线课',
@@ -10,10 +10,11 @@ const title = {
};
module.exports = async (ctx) => {
const caty = ctx.params.caty || 'recommend';
const category = ctx.params.category ?? 'recommend';
const rootUrl = 'https://member.guancha.cn';
const apiUrl = `${rootUrl}/zaixianke/home`;
const response = await got({
method: 'get',
url: apiUrl,
@@ -21,11 +22,11 @@ module.exports = async (ctx) => {
let items;
switch (caty) {
switch (category) {
case 'books':
items = response.data.data.books.map((item) => ({
title: item.title,
link: `${rootUrl}/guanshutang/summary.html?id=${item.id}`,
link: `${rootUrl}/guanshutang/summary.html?id=${item.id}&page=0`,
description: `<img src="${item.cover}"><p>[${item.audio_time}] ${item.desc_short}</p>`,
pubDate: new Date(parseInt(item.cover.split('/').pop().substr(0, 10)) * 1000).toUTCString(),
}));
@@ -53,7 +54,7 @@ module.exports = async (ctx) => {
break;
default:
items = response.data.data[caty].map((item) => {
items = response.data.data[category].map((item) => {
let timeArray = item.media_time && item.media_time.trim().split(/\D+/, 3);
timeArray = timeArray && timeArray.filter((item) => item !== '');
let itunes_duration;
@@ -78,7 +79,7 @@ module.exports = async (ctx) => {
}
ctx.state.data = {
title: `观学院 - ${title[caty]}`,
title: `观学院 - ${titles[category]}`,
link: `${rootUrl}/index.html`,
item: items,
};
lib/routes/guancha/personalpage.js → lib/v2/guancha/personalpage.js
View File
lib/v2/guancha/radar.js
+49
View File
@@ -0,0 +1,49 @@
module.exports = {
'guancha.cn': {
_name: '观察者网',
'.': [
{
title: '头条',
docs: 'https://docs.rsshub.app/new-media.html#guan-cha-zhe-wang-tou-tiao',
source: ['/GuanChaZheTouTiao', '/'],
target: '/guancha/headline',
},
{
title: '首页',
docs: 'https://docs.rsshub.app/new-media.html#guan-cha-zhe-wang-shou-ye',
source: ['/'],
target: '/guancha/:category?',
},
{
title: '观学院',
docs: 'https://docs.rsshub.app/new-media.html#guan-cha-zhe-wang-guan-xue-yuan',
source: ['/'],
target: '/guancha/:category?',
},
],
app: [
{
title: '个人主页文章',
docs: 'https://docs.rsshub.app/new-media.html#guan-cha-zhe-wang-ge-ren-zhu-ye-wen-zhang',
source: ['/user/get-published-list', '/'],
target: (params, url) => `/guancha/personalpage/${new URL(url).searchParams.get('uid')}`,
},
],
member: [
{
title: '观学院',
docs: 'https://docs.rsshub.app/new-media.html#guan-cha-zhe-wang-guan-xue-yuan',
source: ['/'],
target: '/guancha/member/recommend',
},
],
user: [
{
title: '风闻话题',
docs: 'https://docs.rsshub.app/new-media.html#guan-cha-zhe-wang-feng-wen-hua-ti',
source: ['/topic/post-list', '/'],
target: (params, url) => `/guancha/topic/${new URL(url).searchParams.get('topic_id')}/${new URL(url).searchParams.get('order')}`,
},
],
},
};
lib/v2/guancha/router.js
+7
View File
@@ -0,0 +1,7 @@
module.exports = function (router) {
router.get('/headline', require('./headline'));
router.get('/member/:category?', require('./member'));
router.get('/personalpage/:uid', require('./personalpage'));
router.get('/topic/:id/:order?', require('./topic'));
router.get('/:category?', require('./index'));
};
lib/routes/guancha/topic.js → lib/v2/guancha/topic.js
+14 -13
View File
@@ -1,10 +1,10 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const date = require('@/utils/date');
const { parseRelativeDate } = require('@/utils/parse-date');
module.exports = async (ctx) => {
const id = ctx.params.id || '0';
const order = ctx.params.order || '1';
const id = ctx.params.id ?? '0';
const order = ctx.params.order ?? '1';
const rootUrl = 'https://user.guancha.cn';
const currentUrl = `${rootUrl}/${id === '0' ? 'main/index-list.json?' : 'topic/post-list?topic_id=' + id}&page=1&order=${order}`;
@@ -16,28 +16,29 @@ module.exports = async (ctx) => {
const $ = cheerio.load(response.data);
const list = $('.list-item h4 a, ul.home li .list-item h4 a')
.slice(0, 10)
.map((_, item) => {
let items = $('.list-item h4 a, ul.home li .list-item h4 a')
.toArray()
.map((item) => {
item = $(item);
return {
title: item.text(),
link: `${rootUrl}${item.attr('href')}`,
link: `${rootUrl}${item.attr('href')}&page=0`,
};
})
.get();
});
const items = await Promise.all(
list.map((item) =>
items = await Promise.all(
items.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const detailResponse = await got({
method: 'get',
url: item.link,
});
const content = cheerio.load(detailResponse.data);
item.pubDate = date(content('.time1').text(), 8);
item.author = content('.user-main h4 a').eq(0).text();
item.pubDate = parseRelativeDate(content('.time1').text());
item.author = content('.user-main h4 a').first().text();
item.description = content('.article-txt-content').html();
return item;